ESP32 Forum

Hi all,

I have designed a custom PCB which has some I/O s to use. As you can see from the schematics. (C2:1uF, C3:22uF Tantalium)



I have XL4015 as main input power connected (3S9P 18650 LG batteries / 3S40A BMS) to lower it for stable 5.1V (used by RPi) and HT7833 for ESP32 while active mode and deepsleep.
Im measuring the voltage with ADS1115 with a divider 1:10 ratio.
2 output Relays to control on/off RPi and another 12V load (2A max current drawn idle 250mA).

How the system works:

Firstly ESP32 wakes from the deepsleep and measure the battery. Then set Relay1 HIGH to start RPi. Then set Relay2 to start Load.
Waiting for RPi to do its own things and wait for Serial Communication.When RPi asks for current battery level answer it and download the new deep sleep configs.

Wait for RPi to shutdown safely. Set both Relays LOW goes back to sleep.

Where the Problem Starts:

Everything works flawless for the first run. No problem at all. When ESP goes to deep sleep LED on the GPIO13 randomly fades, blinks, flashes..... Just randomly no pattern. I can desolder it but there must be something wrong with it. <br/>

Another and BiGGER problem is the Relay2 (controlling load) randomly clicking. Which is strange because its floating (GPIO25) and LED floating (GPIO13) but Relay1 is stable as rock (GPIO26). Thats a huge problem bc it draws my battery and I cant trust it apperantly.

So there must be something wrong with the design... Something floating somewhere but after 2 weeks for debugging Im lost. <br/>

What I tried:

I have changed ESP32 for at least 20 times with new ones...............Nothing changes. Relay1 still rock solid.
I have changed the Relays with 4 different brand..........................Nothing changes. Relay1 still rock solid.
I have added a Pull-Down resistor with 10K (BC547 base to GND).......Clicking of Relay2 decreases but random blinks on LED increases. Relay1 still rock solid.
I have changed the batteries and switch to DC outlet.....................Nothing changes. Relay1 still working as expected.

I think its not related with the code since I 'm using the same code for breadboard prototype with devkit and its working flawlessly.
Any help would be appriciated.

Seems like something is injecting voltage and current into the circuit, in amounts significant enough to make the LED light up.
Is there an inductive load anywhere near? Transformer, motor, that kind of stuff?
What is the configuration of the GPIOs's pull-ups/pull-downs in deep sleep? Can you enable the pull-downs for deep sleep and see if they sink enough stray current to keep the system stable?
Could there be some unintended conductive path between power and signal lines (board/soldering defect)?
Due to internal ESD protection diodes, it is often problematic to remove power from a chip (RPi SoC,...) when any signal line is externally pulled high.

MicroController wrote: ↑

Fri May 05, 2023 2:43 pm

Seems like something is injecting voltage and current into the circuit, in amounts significant enough to make the LED light up.
Is there an inductive load anywhere near? Transformer, motor, that kind of stuff?
What is the configuration of the GPIOs's pull-ups/pull-downs in deep sleep? Can you enable the pull-downs for deep sleep and see if they sink enough stray current to keep the system stable?
Could there be some unintended conductive path between power and signal lines (board/soldering defect)?
Due to internal ESD protection diodes, it is often problematic to remove power from a chip (RPi SoC,...) when any signal line is externally pulled high.

Yeah after a month, i found it.... BJTs are correct at the schematics but footprints are wrong! Im gonna find the manufacturer and ....